Assigned
Status Update
Comments
kj...@google.com <kj...@google.com> #2
[Empty comment from Monorail migration]
ka...@google.com <ka...@google.com> #4
We are seeing this again on multiple PRs.
For example:
-https://github.com/flutter/flutter/pull/100476
- reports image changes
-https://flutter-gold.skia.org/search?issue=100476&crs=github&patchsets=1
- shows no image changes
-https://flutter-gold.skia.org/json/v1/changelist_summary/github/100476
- the json specifies there are untriaged images
```
{
"changelist_id": "100476",
"patchsets": [
{
"new_images": 1,
"new_untriaged_images": 1,
"total_untriaged_images": 1,
"patchset_id": "7deb72979b845d374955962191c16adbd5dc7b29",
"patchset_order": 1
}
],
"outdated": false
}
```
For example:
-
- reports image changes
-
- shows no image changes
-
- the json specifies there are untriaged images
```
{
"changelist_id": "100476",
"patchsets": [
{
"new_images": 1,
"new_untriaged_images": 1,
"total_untriaged_images": 1,
"patchset_id": "7deb72979b845d374955962191c16adbd5dc7b29",
"patchset_order": 1
}
],
"outdated": false
}
```
kj...@google.com <kj...@google.com> #6
Note for future fix: There was triaging done on the main branch, but not for this CL
is...@google.com <is...@google.com> #7
Edits/notes reflect the following in Monorail: auto-CCs.
kj...@google.com <kj...@google.com> #8
This might still be an issue for the Flutter instance and thus should have some more investigation.
Description
This PR introduced changes to golden file images. The engine roll sheriff triaged them, resulting in no images listed as untriaged in the changelist summary UI (
This change was blocked though by our flutter-gold check that looks up
The result of this at the time was:
```
{
"changelist_id": "93510",
"patchsets": [
{
"new_images": 6,
"new_untriaged_images": 2,
"total_untriaged_images": 2,
"patchset_id": "1097533850dc8ea158597597b2f17dd2339ee0d2",
"patchset_order": 4
}
],
"outdated": false
}
```
So it appeared there were still untriaged images.
Clearing the cache resolved the issue.
I also want to note though that Gold may have been very confused at the time. A new key/value pair was added to our instance and not checked in via pre-submit so Gold was randomly receiving images in post-submit for about a week. This breakdown of process may or may not have caused an issue with the cache,